A Meta-Analysis and Review About the (Un)Fairness Perceptions of Algorithmic Decision-Making
Abstract
This meta-analysis investigates the relationship of algorithmic decision-making (ADM) and perceptions (e.g., fairness) as well as perceived organizational outcomes (e.g., organizational attractiveness), and further explores moderators (cultural cluster, gender, age) to understand algorithm aversion and appreciation within the business and management context. Previous research shows ambiguous results with ADM being perceived as either averse or appreciative. Hence, we propose a synthesis of existing research to shed light on the current status quo. Based on a sample of 21 studies and 67 effect sizes (N = 7,818) we find that, compared to human decision-making, the use of ADM was perceived as less fair and as a bigger threat, and organizations which use ADM were seen as less attractive. Cultural clusters moderated the relationship between perception of ADM and outcomes, whereas the samples’ mean age and percentage of females did not. We developed a framework that qualitatively incorporated further proposed antecedents and contextual factors that affect perceptions of ADM. As algorithm aversion is prevailing, further research needs to be conducted to understand this phenomenon in more detail, so that managers and organizations are able to use ADM most effectively.
Keywords
Algorithmic decision-making; Employee reactions; Applicant reactions; Human resource management; Meta-analysis